Code the Law Weekly #3
Product over Pitches. Repos over Resorts. PRs over PR.
Unpopular Opinion
Product over Pitches. Repos over Resorts. PRs over PR. One of the animating tenets of Code the Law is that an obsession with the superficial, performative aspects of innovation is holding our field back. We tend to overemphasize conferences, flashy invite-only pitch-offs and press mentions, all at the expense of trying, learning, discovering, building, proving and - yes - failing. Too much vanity. Too little curiosity. Let's fix this.
Makers & Doers
David Colarusso teaches how to (carefully) use generative AI to transform a bunch of PDF files into structured, summarized legal data.
The Macrocosm Consortium is a mysterious order of code-wranglers with an absurdly ambitious mission: building digital tools, like the Alexandria Index, with "civilizational scale." Think ALL world religious texts, all patents, all case law, all (open) scholarship.
Solve Intelligence is trying to make patent-writing faster, smarter and cheaper.
Visalaw.Ai teams up with the American Immigration Lawyers Association on a AI-powered tool to help immigration attorneys.
MyPocketCounsel is a Nigerian contract automation startup backed by the Google for Startups Black Founders Fund.
Words Worth Reading This Week
Ilona Logvinova of McKinsey's legal team shares a thoughtful outlook on how lawyers can maximize the benefits of generative AI.
Quentin Steenhuis of Lemma Legal (and Suffolk Law's LIT Lab) explores how LLMs can be used to create sophisticated document automation solutions.
VC firm Andreessen Horowitz is not just funding AI startups, they're helping to create them by curating and sharing knowledge on how to use emerging technology.
MIT PhD/HLS JD student Eric Martinez questions GPT-4's scores on the bar exam.
Keeping an Eye On ... Hyper-Specific Tools
For now, the "know everything, do everything" legal assistants are getting the most attention, and everyone seems to be chasing the dream, or nightmare, of a robot-lawyer-in-a-box, where you tell the machine "do some lawyer stuff" and the machine whirs and whizzes for a few seconds before spitting out fully-baked work product. Like a microwave for legal work.
This vision of legal AI is all about scale and universality: can you sell it to everyone, everywhere, for everything, all at once? Investors love this, because it requires and rewards them.
But maybe the real future of legal AI is micro-scale, not hyper-scale? What if the assistants we need are tiny tools, carefully designed to amplify the intelligence of lawyers operating in specific areas of law? What might it look like for a copyright lawyer to have an AI assistant that's trained specifically to be an expert on fair use? Or for a trial lawyer to have an AI assistant that's tuned specifically on challenging accounting expert witnesses?